-
Vol. 55, No. 3MICROBIOLOGICAL REVIEWS, Sept. 1991, p.
349-3700146-0749/91/030349-22$02.00/0Copyright X) 1991, American
Society for Microbiology
Aromatic Amino Acid Biosynthesis in the Yeast
Saccharomycescerevisiae: a Model System for the Regulation of a
Eukaryotic Biosynthetic PathwayGERHARD H. BRAUS
Mikrobiologisches Institut, Eidgenossische Technische Hochschule
Zurich, CH-8092 Zurich, Switzerland
INTRODUCTION
.......................................................................
349GENE-ENZYME RELATIONSHIPS
...................................................... ...........
.....351
Shikimate
Pathway........................................................................351
AR03 and AR04: DAHP synthase
.............................................
..........................351AROI: arom pentafunctional
enzyme......................................................................352AR02:
chorismate synthase
.......................................................................
353
Phenylalanine-Tyrosine Branch
................................................
.......................353
AR07: chorismate
mutase.......................................... ......353PHA2 and
TYRI: prephenate dehydratase and prephenate
dehydrogenase................................ 353
Tryptophan
Branch.......................................................................354
TRP2 and TRP3C: anthranilate synthase complex .........354
TRP4: phosphoribosyltransferase
.......................................................................354
TRPI: PRA isomerase
.................................................
....................... 354TRP3B: InGP
synthase.......................................................................
355TRPS: tryptophan synthase
........................................................................
355
REGULATION OF ENZYME
SYNTHESIS..............................................355..........355Regulation
of Transcription
.......................................................................
355
Initiation of
transcription.......................................................................
356
(i) A single GCN4-binding site has different functions in the
promoters of theisogenes AR03 and AR04
.....................................3.5......................oo.o...oo
357
(ii) TRP2 and TRP3
promoters.............................................................oo358
(iii) Three GCN4-responsive elements have different functions in
the TRP4 promoter .. ..358
(iv) Two putative GCN4 elements in the TRPS promoter
..............................360
(v) The promoters of AR07, TRPI, and TYR] are not regulated by
GCN4
...............................360mRNAdecay........................................................................
360
Translation.......................................................................360The
transcriptional regulator of amino acid biosynthesis GCN4 is
regulated
at the translational level
.......................................................................
360
REGULATION OF ENZYME ACTIVITY
...........................................................eeo
...361
Phenylalanine-Inhibitable DAHP
Synthase.......................................36.....1........................361The
Allosteric Chorismate Mutase Can Be Locked in the Activated
State................................. 362
Regulation of the Anthranilate Synthase Complex
............................................... .363
CONCLUSIONS
.......................................................................
365ACKNOWLEDGMENTS
..........................................................
.............
365REFERENCES................................................
....................... 365
INTRODUCTION
The biosynthesis of the aromatic amino acids tryptophan(Trp),
phenylalanine (Phe), and tyrosine (Tyr)-especiallythe tryptophan
branch of this pathway-has become one ofthe best-studied examples
of a biosynthetic pathway. Stud-ies of this pathway have
contributed to the understanding oftopics such as gene-enzyme
relationships, promoters, pro-tein-DNA interactions, translational
control, enzyme struc-ture and catalysis, protein-protein
interactions, and controlof flow through a pathway in a wide range
of organisms.Earlier reviews about the aromatic amino acid
biosynthesisoften concentrate on prokaryotic organisms, mainly on
thetryptophan branch, and only a smaller portion also deal
witheukaryotic organisms and includes other parts of the path-way
(5, 6, 37, 46-48, 66, 97, 98, 206, 207). Since theprokaryotic
paradigm of regulatory mechanisms does not
completely extend to eukaryotes, this review focuses on
theregulation of the pathway in a simple eukaryotic system,
theunicellular yeast Saccharomyces cerevisiae. This yeast isone of
the oldest commercially cultured organisms and isalso one of the
best-studied genetic systems available. Sincethe first yeast
transformation (91), yeast research hasboomed and is giving rise to
numerous new insights in theunderstanding of various aspects of a
branched biosyntheticpathway.
Archaebacteria, eubacteria, plants, and fungi are compe-tent to
synthesize de novo the three aromatic amino acidsphenylalanine,
tyrosine, and tryptophan. Animals are gener-ally able to synthesize
tyrosine only by hydroxylation ofphenylalanine and require the
other aromatic amino acids intheir diet (79). Specific inhibitors
of the aromatic pathway,e.g., glyphosate (N-phosphomethylglycine),
can therefore
349
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
0D EGD| AR04 1 [AROI CEDBAI ARI2 COOH
E-4-P CH2+ ~DAHP ~ * Li
PEP OOAR03O XA
0 COOH
II ~~~~~~~HOOC,,,,J
' ' PPA '
II ~~~~~~OH® PHA2 .R1 0
NH2 NH2
II r~~~~COHCOOH
I-----Phe Tyr
OH
.--- - - - - - - - - _
FIG. 1. Biosynthesis of aromatic amino acids and regulation of
the enzymes in S. cerevisiae. The numbers correspond to the
numberingof the enzyme reactions as used in the text.
Abbreviations: CA, chorismate; AA, anthranilate; PPA, prephenic
acid. Enzymes are indicatedby their gene designations: AR031AR04,
DAHP synthases (EC 4.1.2.15); AROIC, DHQ synthase (EC 4.6.1.3);
AROIE, DHQ dehydratase(EC 4.2.1.10); AROID, DHS dehydrogenase (EC
1.1.1.25); AROIB, shikimate kinase (EC 2.7.1.71); AROIA, EPSP
synthase (EC 2.5.1.19);AR02, chorismate synthase (EC 4.6.1.4);
AR07, chorismate mutase (EC 5.4.99.5); PHA2, prephenate dehydratase
(EC 4.2.1.51); TYR],prephenate dehydrogenase (EC 1.3.1.13); TRP2,
anthranilate synthase (EC 4.1.3.27); TRP3C, glutamine
amidotransferase; TRP4, anthra-nilate phosphoribosyltransferase (EC
2.4.2.18); TRPI, PRA isomerase; TRP3B, InGP synthase (EC 4.1.1.48);
TRP5, tryptophan synthase (EC4.2.1.20).
be used as herbicides and are also inhibitors of microbialgrowth
(115).The seven enzyme-catalyzed reactions of the shikimate
pathway from erythrose 4-phosphate (E4P) and
phospho-enolpyruvate (PEP) to chorismic acid are common for
allaromatic amino acids. The series of reactions is invariable
inall eukaryotic and prokaryotic organisms studied so far
(81).Chorismic acid is the last common intermediate of the
threearomatic amino acids and is distributed towards
tryptophan,phenylalanine/tyrosine, and derivatives therefrom such
asvitamin K, ubiquinone and p-aminobenzoate. Whereas
thebiosynthesis of tryptophan from chorismic acid proceeds infive
invariable steps in all organisms so far studied, twoseparate
routes of phenylalanine and tyrosine biosynthesisexist. Thus,
phenylalanine may be formed from arogenate orfrom phenylpyruvate,
whereas tyrosine synthesis may pro-ceed from either arogenate or
4-hydroxyphenylpyruvate. InS. cerevisiae only the phenylpyruvate
and the 4-hydroxy-phenylpyruvate pathways have been found (81).An
outline of the biosynthetic pathway in S. cerevisiae
from E4P and PEP through chorismate to the aromaticamino acids
and the other metabolically important com-pounds is given in Fig.
1.The aromatic amino acids are energetically the most costly
amino acids for the living cell: 78 mol of ATP is required
tosynthesize 1 mol of tryptophan; the respective values for
phenylalanine and tyrosine are 65 and 62 mol. On averagethis is
approximately twice the energy required for any otheramino acid
(12). Accordingly, the concentration of theseamino acids in the
cell is among the lowest of all amino acids:in S. cerevisiae the
total pool of phenylalanine, tyrosine, andtryptophan was determined
as 0.6, 0.5 and 0.02 mM, respec-tively (61, 101).Although the
enzymatic steps involved in aromatic amino
acid biosynthesis are very similar in all species studied so
far(79), there are striking differences among various species inthe
genetic organization of the enzyme activities that cata-lyze the
reactions and in their regulation (46, 203). Forexample, in the
enteric bacterium Escherichia coli, all genesof the tryptophan
braneh of the pathway are arranged in thewell-studied tryptophan
operon that permits simultaneousregulation of gene expression by
repression and attenuation(206). In contrast, the tryptophan genes
are scattered overthe genome in all eukaryotic organisms studied to
date, andtherefore each of them requires its own regulatory
signalsequences (98).Some of the encoded enzymes appear to be more
highly
organized in eukaryotic than in prokaryotic
microorganisms.Different fusion patterns have produced
multifunctional en-zymes with different combinations of activity
domains (98).In the eukaryotic microorganisms studied, the activity
do-mains are located on fewer polypeptide chains encoded by
350 BRAUS
I
--------------------
- - - - -
III
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 351
TABLE 1. Genes and enzymes for the biosynthesis of chorismate in
S. cerevisiae
Chromo- mRN Polypeptide sizeReaction Compound Enzyme designation
Gene some (eNkb)dasz
PEP + E4P1 4 DAHP synthase (EC 4.1.2.15) AR03 IV 1.2 41 (370
aa)
AR04 II 1.4 39 (367 aa)DAHP
2 4 DHQ synthase (EC 4.6.1.3) AROIC IV 175 (1,588 aa): aa
1-392DHQ
3 4 DHQ dehydratase (EC 4.2.1.10) AROIE IV aa 1059-1293DHS
4 4 DHS dehydrogenase (EC 1.1.1.25) AROID IV aa
1306-1588Shikimate
5 4 Shikimate kinase (EC 2.7.1.71) AROIB IV aa 886-1059Shikimate
3-phosphate
6 4 EPSP synthase (EC 2.5.1.19) AROIA IV aa 404-886EPSP
7 4 Chorismate synthase (EC 4.6.1.4) AR02 VII 1.4Chorismate
IPhe, Tyr, Trp, ubiquinone,p-aminobenzoate, vitamin K
aaa, amino acid.
fewer genes than in most prokaryotes. An impressive exam-ple is
the pentafunctional arom enzyme, which is found innumerous lower
eukaryotes and which catalyzes reactions 2to 6 of the shikimate
pathway (Fig. 1). In S. cerevisiae thearom enzyme is encoded by the
AROJ gene (58, 114). Incontrast, the genes encoding the
corresponding activities ofE. coli are widely scattered about the
genome, encoding fiveseparable enzymes (159). The diversity in the
patterns ofgene and enzyme organization found in different species
is aremarkable feature of the arom system (203).The aromatic amino
acid biosynthesis in S. cerevisiae is
controlled by two principal mechanisms: (i) regulation ofenzyme
synthesis by the regulation of gene expression, and(ii) regulation
of the enzyme activities that control thecarbon flow.
(i) At the transcriptional level, most of the structural genesof
the aromatic amino acid biosynthesis in S. cerevisiae areregulated
by the transcriptional activator GCN4 (10, 83, 87,94, 191). The
GCN4 protein is the regulator of a complexregulatory network, known
as the general amino acid con-trol, which couples transcriptional
derepression of at least 30structural genes involved in multiple
amino acid biosyntheticpathways (87, 144, 172). (ii) At the enzyme
level the carbonflow is controlled mainly by modulating the enzyme
activi-ties at the first step and at the branch points. In general,
theend products of the major terminal pathways,
phenylalanine,tyrosine, and tryptophan, serve as sensors to control
carbonflow (Fig. 1).
GENE-ENZYME RELATIONSHIPS
In S. cerevisiae, 12 genes encoding enzymes for 15 of the17
reactions in the biosynthesis of the three aromatic aminoacids have
been described (Fig. 1; Tables 1 to 3). Thenumber of genes encoding
the aminotransferases, whichcatalyze the final steps in the
phenylalanine and tyrosinebranches, is as yet unknown.
Shikimate Pathway
The seven enzyme-catalyzed reactions of the commonshikimate
pathway leading to the branch point compoundchorismic acid are
encoded by four genes. Table 1 summa-rizes some features of the
genes and enzymes involved in thebiosynthesis of chorismate in S.
cerevisiae.ARO3 and ARO4: DAHP synthase. 3-Deoxy-D-arabinohep-
tulosonate 7-phosphate (DAHP) synthase (EC 4.1.2.15) car-ries
out the initial step in the shikimate pathway, which is
thecondensation of PEP and E4P to form DAHP in a reactionclosely
analogous to an aldol condensation (for a review, seereference
79).
In the yeast S. cerevisiae, two isoenzymes of DAHPsynthase
exist, one of which is feedback inhibitable byL-phenylalanine and
the other by L-tyrosine (117). Otherorganisms such as E. coli (36)
or the filamentous fungusNeurospora crassa (145) possess three DAHP
synthases,each one regulated by one of the three aromatic amino
acids.Meuris et al. (127) and Teshiba et al. (190) isolated aro3
andaro4 mutations bearing deficiencies in the tyrosine-
andphenylalanine-sensitive DAHP synthases,' respectively.
Thecorresponding genes, AR03 and 'ARO4, are located ondifferent
chromosomes (AR03 on chromosome IV andAR04 on chromosome II)
(llla). The two genes werecloned, and they encode a 1.2- and a
1.4-kb mRNA, respec-tively (llla, 154, 155, 190). From the AR03 DNA
sequenceone can predict a protein of 370 amino acids with a
calcu-lated molecular mass of 41 kDa (154); the AR04
sequence'predicts a polypeptide of 367 amino acids with a
molecularmass of 39 kDa (llla). The amino acid sequences of the
twogenes show strong overall similarity (75% according to themethod
proposed by Gribskov and Burgess [72]) including225 identical amino
acid residues. A high degree of similarity(65 to 71%) is also found
with the three DAHP synthases ofE. coli (52, 174, 214).The
ARO3-encoded enzyme was purified to apparent
homogeneity and has a molecular mass of 42 kDa, corre-
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
TABLE 2. Genes and enzymes for the biosynthesis of phenylalanine
and tyrosine in S. cerevisiae
mRNA Polypeptide sizeReaction Compound Enzyme designation Gene
Chromosome length (kDa)a
(kb)
Chorismate8 4 Chorismate mutase (EC 5.4.99.5) AR07 XVI 0.95 30
(256 aa)
Prephenate9 4 Prephenate dehydratase (EC 4.2.1.51) PHA2 XIV
Phenylpyruvate10 4 Phenylalanine aminotransferase
Phenylalanine
Chorismate8 4
Prephenate11 4 Prephenate dehydrogenase (EC 1.3.1.13) TYR] II 52
(441 aa)
4-Hydroxyphenylpyruvate12 4 Tyrosine aminotransferase
Tyrosine
aaa, amino acid.
sponding to the predicted molecular mass deduced from the lyzes
the second reaction in chorismate biosynthesis, whichDNA sequence
(156) (see below). results in cyclic DHQ after removal of a
phosphate and anAROI: arom pentafunctional enzyme. In S. cerevisiae
the internal oxidation reaction. In an alignment of the DNA-
central five steps of the shikimate pathway (reactions 2 to 6
derived protein sequence, the first 392 amino acid residues ofin
Fig. 1 and Table 1) are catalyzed by a pentafunctional the AROI
gene are similar to the E. coli aroB-encoded DHQenzyme, the arom
multifunctional enzyme, which is encoded synthase (129). There is
36% identity between the twoby the ARO1 gene located on chromosome
IV (54, 114, 133). sequences; including two subdomains of greater
similarityThe protein sequence deduced from the DNA sequence
(58).corresponds to a polypeptide of 1,588 amino acids with a The
3-dehydroquinate (DHS) dehydratase (EC 4.2.1.10)calculated
molecular mass of 175 kDa (58). The yeast AROI catalyzes the
reaction that converts DHQ into DHS andDNA fragment also
complements the corresponding aroA, introduces the first double
bond of the aromatic ring (reac-aroB, aroD, and aroE mutants of E.
coli (114). Functional tion 3). Twenty-one percent of amino acids
1059 to 1293 ofregions within the polypeptide chain have been
identified the arom enzyme are identical to amino acids in the
corre-by comparison with the sequences of the five separate
sponding E. coli aroD gene product (57). Confirmation
thatmonofunctional E. coli enzymes whose activities correspond this
region of the S. cerevisiae sequence truly encodes theto those of
the arom multifunctional enzyme (159). Accord- DHQ dehydratase
activity is provided by the presence of aingly, the pentafunctional
arom enzyme is a mosaic of pentadecapeptide of the corresponding N.
crassa enzymemonofunctional domains connected by some extra amino
which is part of the active site of the enzyme (58).acid residues
as linkers. The arrangement of the domains in Dehydroshikimate is
converted to shikimate by the dehy-the corresponding AROJ gene does
not correlate with the droshikimate (DHS) dehydrogenase (EC
1.1.1.25) catalyzingsuccession of the corresponding catalyzed
reactions in the the fourth step of the pathway. Shikimic acid gave
its namepathway (58) (Table 1). to the pathway and was first
described as a natural productThe 5-dehydroquinate (DHQ) synthase
(EC 4.6.1.3) cata- from the plant Illicium religiosum. It was from
the Japanese
TABLE 3. Genes and enzymes for the biosynthesis of tryptophan in
S. cerevisiae
Reaction Compound Enzyme designation Gene Chromosome
lenPop(kDa)(kb)
Chorismate + glutamine13 4 Anthranilate synthase (EC 4.1.3.27)
TRP2 V 1.8 60 (528 aa)
(glutamine amidotransferase) TRP3C XI 1.75 54 (484 aa): aa
1-206Anthranilate + PRPP
14 4 Anthranilate phosphoribosyltransferase TRP4 IV 1.4 41 (380
aa)(EC 2.4.2.18)
PRA15 4 PRA isomerase TRPI IV 0.8-1.0 28 (224 aa)
CDRP16 4 InGP synthase (EC 4.1.1.48) TRP3B XI 1.75 54 (484 aa):
aa 218-484
InGP + serine17 4 Tryptophan synthase (EC 4.2.1.20) TRP5 VII 77
(707 aa)
Tryptophan
aaa, amino acid.
352 BRAUS MICROBIOL. REV.
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 353
name of this plant, shikimi-no-ki, that the name shikimic
acidwas derived (79). In the early 1950s, however, it wasrevealed
that shikimic acid was an obligatory intermediate inthe pathway
from carbohydrate to the aromatic amino acids(53, 177).The
C-terminal domain of the arom polypeptide (residues
1306 to 1588) is similar to the E. coli aroE gene product
DHSdehydrogenase (9), with a 25% identity of amino acids.The
shikimate kinase (EC 2.7.1.71) catalyzes the fifth
reaction, the formation of shikimate-3-phosphate from shiki-mate
and ATP. A similarity of 23% with the E. coli aroLgene product
(130) extends from residues 886 to 1059 on thearom peptide. There
is one well-conserved region betweenresidues 895 and 909
corresponding to the ATP-binding siteof various enzymes.The
5-enolpyruvylshikimate 3-phosphate (EPSP) synthase
(EC 2.5.1.19) condenses shikimate-3-phosphate and a sec-ond
molecule of PEP to produce EPSP (reaction 6). Thisenzyme is the
target of the commercially important herbicideglyphosate
(N-phosphomethylglycine) (8), which is widelyused as a nonselective
herbicide and, in addition, is aninhibitor of microbial growth
(115). The finding that glypho-sate is antagonized by one or more
of the aromatic aminoacids is true for many organisms including
prokaryotes,algae, and plants (71). The yeast EPSP synthase domain
islocated between residues 404 and 866 and is the bestconserved of
the five arom domains when compared with thecorresponding E. coli
domains. It shares an overall 38%identity with the E. coli aroA
gene product (59), with twosubdomains of higher similarity
separated by a region withno similarity.
Similar arom enzymes to those in S. cerevisiae seem to
berestricted to the fungi and the euglenoids (6), whereas inplants
and bacteria several separable enzymes encoded bygenes in different
arrangements have been found (203).Besides S. cerevisiae,
arom-encoding genes have been iso-lated from other ascomycetes such
as the yeast Schizosac-charomyces pombe (141) and Aspergillus
nidulans (40). TheN. crassa enzyme was shown to be a dimer
consisting of twoidentical pentafunctional polypeptides (45, 68,
113).AR02: chorismate synthase. Finally, chorismate is gener-
ated in reaction 7 by removal of a phosphate and introduc-tion
of a second double bond by chorismate synthase (EC4.6.1.4). In S.
cerevisiae the enzyme is encoded by theAR02 gene (54), located on
chromosome VII (133). The genewas recently cloned and codes for a
1.4-kb mRNA (lOOa).
Phenylalanine-Tyrosine Branch
Chorismic acid is the last common intermediate of thethree
aromatic amino acids and is distributed toward
thephenylalanine-tyrosine and the tryptophan branches. In
ad-dition, the chorismate pool in the cell is necessary for
thesynthesis of other aromatic compounds such as vitamin
K,ubiquinone, or p-aminobenzoate. Synthesis of these com-pounds
will not be discussed. In the phenylalanine-tyrosinesine branch of
the pathway, chorismate is converted toprephenate, which is the
last common intermediate beforethe pathway branches again toward
either phenylalanine ortyrosine (Fig. 1). The
phenylalanine-tyrosine branch in-cludes five enzyme reactions. The
genes for three of thesereactions have been identified. Since no
mutants for trans-amination of tyrosine or phenylalanine were
found, thenumber of aminotransferases catalyzing the final step
inthese two branches is unknown. Table 2 summarizes some
features of the genes and enzymes involved in the biosyn-thesis
of phenylalanine and tyrosine in S. cerevisiae.AR07: chorismate
mutase. The first step in the phenylala-
nine-tyrosine branch which is still common in all
organismsstudied so far is the intramolecular rearrangement of
theenolpyruvyl side chain of chorismate to yield
prephenate(reaction 8). The reaction is formally analogous to a
Claisenrearrangement and is catalyzed by chorismate mutase
(EC5.4.99.5).The S. cerevisiae AR07 gene located on chromosome
XVI
(133) encodes a monofunctional chorismate mutase (110,170), a
situation also found in other yeasts (21), in differentplants (67),
and in bacteria such as Bacillus subtilis Marburg(118) and
Streptomyces aureofaciens (69, 70). The yeastAR07 gene was cloned
(13, 170) and was shown to beidentical to a gene necessary for
growth in hypertonicmedium, OSM2 (13). The reason for this
connection betweenAR07 and osmotic stability is unclear.AR07
encodes a 0.95-kb mRNA. DNA sequencing deter-
mined a 771-bp open reading frame (ORF) capable of encod-ing a
protein of 256 amino acids (170). The protein waspurified (168),
and the monomer size of 30 kDa correspondsto the predicted size
deduced from the DNA sequence.The yeast chorismate mutase is not
only feedback inhib-
ited by tyrosine, one of the two end products of
thisbiosynthetic branch, but also strongly activated by tryp-tophan
(110), the end product of the other branch. Theregulation of the
enzyme is discussed in more detail below.The monofunctional B.
subtilis Marburg chorismate mutaseis inhibited by prephenate but
unaffected by tyrosine, phen-ylalanine, or tryptophan (119), and
the S. aureofaciensenzyme activity is unregulated (70). Other
investigated or-ganisms, such as E. coli, use two bifunctional
enzymes: achorismate mutase-prephenate dehydratase (pheA) that
isfeedback inhibited by phenylalanine and a chorismate
mu-tase-prephenate dehydrogenase (tyrA) that is feedback inhib-ited
by tyrosine (50, 51). In both cases the N-terminal part ofthe
bifunctional enzyme carries the chorismate mutase ac-tivity (96,
122). In contrast to other enzymes in this pathway,no significant
similarity between the monofunctional yeastchorismate mutase and
the corresponding domains of thetwo bifunctional E. coli enzymes
was found (170).PHA2 and TYR): prephenate dehydratase and
prephenate
dehydrogenase. For the biosynthesis of phenylalanine
andtyrosine, two alternative routes exist in nature, a
phe-nylpyruvate and a 4-hydroxyphenylpyruvate route, respec-tively,
or an L-arogenate route. Virtually every conceivablecombination of
possible enzyme, arrangements has beenfound: whereas S. cerevisiae
or E. coli use only the phe-nylpyruvate-4-hydroxyphenylpyruvate
routings (116, 117,159) (Fig. 1), plants utilize arogenate as the
sole precursor ofboth phenylalanine and tyrosine (23). A widespread
combi-nation, e.g., in cyanobacteria, is an
arogenate-to-tyrosine/phenylpyruvate-to-phenylalanine pathway. In
other bacte-ria, e.g., Pseudomonas aeruginosa, the two
alternativepathways to phenylalanine and/or tyrosine coexist
(81).
Little is known about the final phenylalanine branch in
S.cerevisiae. Prephenate dehydratase (EC 4.2.1.51) catalyzesthe
first reaction, the conversion of prephenate to phe-nylpyruvate
(reaction 9). Lingens et al. (116) isolated mu-tants with mutations
in the prephenate dehydratase-encod-ing gene PHA2, which is located
on chromosome XIV (133).No mutants have been isolated for the final
reaction (reac-tion 10), the transamination of phenylpyruvate to
phenylal-anine by an aminotransferase. This might be explained
bythe finding that in other organisms there are numerous
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
aminotransferases, often exhibiting a rather broad range
ofsubstrate specificities (159).The yeast TYR] gene (116) in the
final tyrosine branch
encodes prephenate dehydrogenase (EC 1.3.1.13), whichcatalyzes
the oxidative decarboxylation and dehydratationof prephenate
(reaction 11) that results in 4-hydroxyphe-nylpyruvate. The gene is
localized on chromosome 11 (133).The TYR] gene was cloned and
contains an ORF of 441codons for a calculated protein of 52 kDa.
There is aconsensus sequence for an NAD-binding site within the
first45 amino acids that is typical for dehydrogenases (121). As
inthe phenylalanine branch, the aminotransferase(s) involvedin the
final transamination of 4-hydroxyphenylpyruvate totyrosine
(reaction 12) has not yet been characterized.
Tryptophan Branch
The tryptophan branch proceeds in five steps from cho-rismate to
tryptophan, using a set of seven enzyme activitydomains encoded by
five genes in S. cerevisiae. The activi-ties are organized into
four separable enzyme components.Reactions 13 and 16 are catalyzed
by a bifunctional complexcomposed of two subunits encoded by the
genes TRP2 andTRP3. Table 3 summarizes some features of the genes
andenzymes involved in the biosynthesis of tryptophan in
S.cerevisiae.TRP2 and TRP3C: anthranilate synthase complex. The
first
step of the tryptophan branch (reaction 13) is the conversionof
chorismate to anthranilate with glutamine as the donor ofthe amino
group. The enzyme that catalyzes this reaction iscalled
anthranilate synthase (EC 4.1.3.27) and is feedbackinhibitable by
the end product of the branch, tryptophan.Two genes, TRP2 and TRP3,
located on chromosome V andXI, respectively, are necessary to
encode the enzyme. Bothgenes were cloned (3) and sequenced (212).
TRP2 and TRP3encode 1.8- and 1.75-kb mRNAs (28, 212) with open
readingframes of 528 and 454 codons for calculated polypeptides
of60 and 54 kDa, respectively (212). Purification of the
enzymecomplex confirmed these data and revealed a molecularmass for
the subunits of 64 and 58 kDa, respectively (160)(see below).The
TRP2 gene product encodes an anthranilate synthase
activity that is able to form anthranilate, with
considerablyreduced efficiency, only if provided with ammonia
instead ofglutamine (Gln) (160). It also contains the
tryptophan-bind-ing site for feedback inhibition. An amino acid
sequencealignment between the S. cerevisiae TRP2 gene and
corre-sponding E. coli trpE gene (142) exhibits only a limitedamino
acid sequence similarity: nine short conserved regionscan be found,
with eight of them located in the C-terminalhalf of the enzyme. A
comparison of 16 amino acid sequencesderived from the corresponding
nucleotide sequences of dif-ferent species, namely 12 anthranilate
synthases and 4 se-quences of a similar enzyme, p-aminobenzoate
synthase, alsoshows high variability in the amino-terminal half of
themolecule and conserved regions in the distal part of themolecule
(48).The TRP3 gene encodes two enzyme activities. The first of
these, encoded by the N-terminal part of the enzyme, is
aglutamine amidotransferase activity, which provides thenitrogen
from glutamine for the synthesis of anthranilate (2,212). An
alignment of the first 206 amino acids of the TRP3product with 195
amino acids of the E. coli trpG product(208) shows 38% identity
(212). There is a 60% identity withthe corresponding gene product
of N. crassa (167). Similarvalues also have been found with four
other known fungal
sequences, including A. nidulans, A. niger,
Penicilliumchrysogenum, and Phycomyces blakesleeanus (48). A
74%identity exists with the corresponding gene product of theyeast
Hansenula polymorpha (163). Whereas the E. colitrpG gene is fused
to the gene encoding anthranilate phos-phoribosyltransferase, which
catalyzes the next reaction(reaction 15), the yeast and other
fungal glutamine amido-transferases are fused to the
indole-3-glycerol-phosphate(InGP) synthase catalyzing the fourth
step of the tryptophanbranch (see the section TRP3B: InGP Synthase,
below).
TRP4: phosphoribosyltransferase. The second reaction ofthe
tryptophan branch is the transfer of a 5-phosphoribosylmoiety from
5-phosphoribosylpyrophosphate to the aminogroup of anthranilate,
resulting in phosphoribosylanthra-nilate (PRA). The reaction is
catalyzed in S. cerevisiae by amonofunctional anthranilate
phosphoribosyltransferase (EC2.4.2.18) encoded by the TRP4 gene.
The TRP4 gene, whichis located on chromosome IV, was cloned and
encodes al.A-kb mRNA (63, 64) which contains an ORF of 380
codonsfor a putative protein of 41 kDa (64). Purification of
theenzyme revealed a monomer size of 42 kDa on a
denaturingpolyacrylamide gel (93).The product of the corresponding
E. coli gene, trpD, the
C-terminal part of a fused trpG-D gene (208), is only
partiallysimilar to the yeast protein, with 15% overall
identity;however, 50 and 44% identities exist in two separate
do-mains of about 50 amino acids each (64, 98). No other
fungalamino acid sequences are as yet available for
comparison.TRPI: PRA isomerase. PRA isomerase catalyzes a
practi-
cally irreversible Amadori rearrangement, the third
step(reaction 15) in the tryptophan branch of the pathway.
Theaminoglycoside PRA undergoes an internal redox reaction,which
results in the ketone carboxyphenylamino-l-deoxy-ribulose
5-phosphate (CDRP).The corresponding gene in S. cerevisiae is the
TRPI gene
located on chromosome IV. The yeast TRPI gene was one ofthe
early yeast genes that was cloned by complementation ofthe
corresponding E. coli mutant (180, 188). The geneattracted special
interest, because an ARS (autonomousreplication sequence) site is
located adjacent to the 3' end ofthe TRPI gene, which allows the
use of the yeast TRPI-ARSfragment as a selectable marker in many
extrachromosom-ally maintained yeast vectors. The TRPI gene encodes
aheterogeneous mRNA of 0.8 to 1.0 kb (27, 28) with an ORFof 224
codons for a calculated protein of 28 kDa (194).Purification of the
enzyme (28) revealed a molecular mass of23 kDa for the
protein.Whereas S. cerevisiae exhibits a monofunctional PRA
isomerase, the situation is different in other
ascomycetes:several genes have been cloned that encode a
trifunctionalpolypeptide with the arrangement NH2-glutamine
amido-transferase-InGP synthase-PRA isomerase-COOH. Geneswith this
arrangement have been cloned from different fungiincluding N.
crassa (167), A. nidulans (137), A. niger (105),Cochliobolus
heterostrophus (195), Penicillium chrysoge-num (166), and
Phycomyces blakesleeanus (164). For thebasidiomycete Schizophyllum
commune a bifunctional NH2-InGP synthase-PRA isomerase-COOH was
proposed (139).In S. cerevisiae two genes encode these three
enzymaticfunctions: TRP3 encodes the bifunctional
NH2-glutamineamidotransferase-InGP synthase-COOH, and TRPI
encodesthe monofunctional PRA isomerase. The same pattern oftwo
separated genes was found only in a series of Saccha-romyces
strains (25), in Kluyveromyces lactis (178), and inCandida maltosa
(20). A single trifunctional gene seems tobe present in Hansenula
spp. and S. pombe (18, 27a, 192).
354 BRAUS
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 355
The apparent restriction of this arrangement to a small groupof
fungi suggests that the S. cerevisiae TRPI gene may havebeen
separated from a trifunctional gene found in otherascomycetes by a
translocation that occurred relativelyrecently in the evolution of
yeasts. More argument for thishypothesis are summarized in
Regulation of the AnthranilateSynthase Complex (below).
In many bacteria including E. coli, the PRA isomerase isfused to
the InGP synthase. A monofunctional PRA isomer-ase, as in S.
cerevisiae, was found in fluorescent pseu-domonads. There is a 25%
identity between the amino acidsequences of the E. coli and S.
cerevisiae enzymes, a 40 to45% identity with those of the various
filamentous fungi, anda 53% identity between those of S. cerevisiae
and K. lactis(178). The last value is low if one considers that
both aresimilar species of budding yeasts and that other
genesanalyzed so far (e.g., the two URA3 genes encoding
oroti-dine-5'-phosphate decarboxylase) share a much higher
pro-portion (80%) of identical amino acids (178).The crystal
structure of the fused E. coli protein NH2-
InGP synthase-PRA isomerase-COOH has been solved(162): both
domains are eightfold a$ barrels resemblingtriose phosphate
isomerase (TIM barrel). Priestle et al. (162)aligned all known
sequences and demonstrated that thepredicted a-helices, turns, and
13-strands were coincidentwith the known ones of E. coli. Luger et
al. (120) demon-strated, by using mutated yeast TRPJ genes encoding
circu-larly permutated variants, that the yeast PRA isomerase
alsofolds in a TIM barrel.TRP3B: InGP synthase. The fourth step of
the tryptophan
pathway (reaction 16) is the decarboxylation of CDRP andthe
closure of the second ring to yield InGP. The reaction iscatalyzed
by InGP synthase (EC 4.1.1.48), the second do-main of a
bifunctional enzyme which is encoded in S.cerevisiae by the 3' half
of the TRP3 gene (TRP3B).Sequence alignment reveals that codons 218
to 484 of the
ORF on the 1.75-kb TRP3 mRNA correspond to this do-main. The
degree of identity to the E. coli domain trpC (208)is 32% (212) and
is therefore higher in comparison with thePRA isomerase and lower
in comparison with the glutamineamidotransferase of E. coli. A
multisequence alignmentreveals that all InGP synthases are more
highly conservedthan the PRA isomerases and also fold in a TIM
barrel. Themain areas of conservation are located in the ,B-strands
of thebarrel, and some are located in the turns at the carboxyl
endsof the strands, whereas the a-helical regions seem
generallymore variable (48, 162).TRPS: tryptophan synthase. In the
final reaction of the
tryptophan branch (reaction 17) the InGP is cleaved andindole is
condensed with serine to yield tryptophan. E. colitryptophan
synthase (EC 4.2.1.20) is one of the most inten-sively studied
enzymes of the pathway. The enzyme has twoactive sites, one for the
aldol cleavage of InGP to yieldindole and
glyceraldehyde-3-phosphate, and the other for thesynthesis of
L-tryptophan from indole and serine (for areview, see reference
128). In most organisms both functionsare on two separate
polypeptide chains. In S. cerevisiae asingle gene, TRP5, located on
chromosome VII (133) en-codes the bifunctional tryptophan synthase
protein with adeduced amino acid sequence of 707 amino acids and
acalculated molecular mass of 77 kDa (213). Purification ofthe
enzyme reveals a size for the monomer of 76 kDa (55).The N-terminal
domain of 239 amino acids of the yeast
enzyme is similar to the E. coli a-subunit (29% identity);
thedistal 389 amino acids correspond to the ,B-subunit
(50%identity). This order of segments is the reverse of the
chromosomal order characteristic of all prokaryotes thathave
been examined. A single tryptophan synthase with thesame gene
domain order as in S. cerevisiae was also found inthe filamentous
fungus N. crassa (34). The two fungalenzymes show strong similarity
when compared on thededuced amino acid sequence level: the A
domains have54% identity; the B domains have 75% identity (34).
Analignment of known DNA sequences suggests that the
basicthree-dimensional structure is probably the same whetherthe
subunits are fused or not (49). It is known from thecrystal
structure that the a-subunit is in the form of a TIMbarrel, as are
the PRA isomerase and the InGP synthase (99).The 45-amino-acid
connector region of S. cerevisiae has lessthan 25% identity to the
54 amino acids of N. crassa,although secondary-structure analysis
predicts that bothconnectors would be a-helical.
REGULATION OF ENZYME SYNTHESIS
The amount of a certain enzyme in a cell is determined bythe
rate of protein synthesis and degradation. Protein syn-thesis is
determined by gene expression, which includesvarious parameters
such as the initiation, elongation, andtermination of
transcription; the capping, processing, andpolyadenylation of the
transcript; the packaging into ribonu-cleoprotein particles; the
transport of the mRNA from thenucleus into the cytoplasm; and,
finally, the initiation, elon-gation, and termination of
translation.Not much is known about the regulation of protein
degra-
dation of the enzymes of the aromatic amino acid biosynthe-sis.
There is no evidence, however, that there are
significantdifferences in the degradation rates of the different
enzymes.The regulation of enzyme synthesis of the aromatic
amino
acid biosynthetic genes takes place mainly at the level
oftranscription and specifically at the initiation of
transcrip-tion. Additional regulatory points are the mRNA
half-lifeand translational control of the level of the main
transcrip-tional regulator, the protein GCN4.
Regulation of Transcription
Transcription of the genes of the aromatic amino
acidbiosynthesis is regulated mainly at the 5' end of the
genes,where DNA-binding proteins determine the rate of initiationof
transcription at the different promoters. As with most ofthe yeast
genes, the aromatic amino acid biosynthetic genescarry no introns
and hence cannot be regulated by splicing(62). As in higher
eukaryotic cells, the mature yeast mRNAspossess poly(A) tails,
which seem to be either the product ofprocessing and
polyadenylation or a coupling of terminationand polyadenylation
(35, 149, 150). Besides the initiation oftranscription, the decay
rate of the mRNA can influencegene expression (29).The
transcriptional regulation of the amino acid biosyn-
thetic genes in the yeast S. cerevisiae includes three
impor-tant features which are different from those of the
corre-sponding genes in a bacterium such as E. coli. (i) The
yeastgenes involved in amino acid biosynthesis are spread all
overthe genome (133) and are not organized in operons, as is
thecase for some biosynthetic pathways of E. coli. Therefore,the
expression of all genes takes place independently and theinitiation
of transcription is performed on individual promot-ers. (ii) Yeast
cells maintain a significant level of amino acidbiosynthetic gene
expression when amino acids are added tothe growth medium or when
large internal amino acid poolsare present (11, 132). This
relatively high level of transcrip-
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
tion-and, as a consequence, of enzyme synthesis-in thepresence
of amino acid excess has been named the basal-level control (11).
Therefore, unlike bacteria, yeasts haverelatively large
intracellular amino acid pools (61, 101). (iii)Yeast cells respond
to starvation for a single amino acid byturning on the
transcription of at least 30 genes in unrelatedamino acid
biosynthetic pathways. For example, starvationfor tryptophan leads
not only to the derepression of theenzymes of the aromatic amino
acid biosynthetic pathway,but also to the biosynthetic enzymes of
arginine, histidine,isoleucine, leucine, lysine, serine, and
threonine, and todifferent aminoacyl-tRNA synthetases. Therefore,
the path-way of aromatic amino acid biosynthesis in S. cerevisiae
ispart of a complex regulatory network known as generalcontrol (87,
172). This cross-pathway regulation also existsin numerous other
yeasts (21) and in other fungi such as N.crassa and A. nidulans
(38, 158), enhancing the potential ofthe organisms to survive in an
environment of externalamino acid imbalance (144).
In many bacteria, including E. coli, transcription of thegenes
for their amino acid biosynthetic enzymes is repressedwhen the
amino acids are present in sufficient amounts in thegrowth medium.
Starvation for a single amino acid leadsmainly to increased
transcription of only the genes in thecorresponding pathway. An
additional control system,termed metabolic regulation, has been
described for thearomatic amino acid biosynthesis in E. coli and
seems to beindependent of the presence or absence of these amino
acids(159).
In S. cerevisiae at least six amino acid biosyntheticpathways,
namely those for arginine, lysine, methionine,leucine, isoleucine,
and valine, are, independently of or inaddition to the general
control system, also controlled byspecific regulatory mechanisms
(125). With the exception ofarginine, these specific control
mechanisms seem to operateat the transcriptional level. For the
aromatic amino acidbiosynthetic enzymes, however, no specific
regulatory sys-tem has been found.
Initiation of transcription. During the last couple of years
ithas been demonstrated that the mechanisms necessary forthe
initiation of transcription at RNA polymerase II promot-ers are in
principle conserved between yeasts and humans.Yeast transcriptional
activators often have a related coun-terpart in other organisms,
with a high degree of similarity infunctionally important domains.
In some cases it has beenshown that the related proteins of the
higher organism areable to complement defects in the corresponding
yeast gene.In addition, some of these proteins have been shown to
beoncogene products in the higher organism.One example is the
regulator protein GCN4 of S. cerevi-
siae, which is required for the response to amino
acidstarvation. GCN4 shares homology with thejun oncoproteinand the
human trans-activator protein AP-1. The GCN4DNA-binding domain can
be exchanged for the jun DNA-binding domain, and the resulting
chimeric protein is stillactive in S. cerevisiae (22, 184). GCN4
contains the leucinezipper structure responsible for dimerization,
which is acharacteristic feature of a whole class of
DNA-bindingproteins (4). GCN4 activates transcription in the
generalcontrol system of the amino acid biosynthesis network of
S.cerevisiae (89). As a result, derepressed specific enzymelevels
of the gene products of the corresponding regulatedgenes are
measured (132).
Typical amino acid biosynthetic promoters are dual pro-moters
and hence can be regulated by two control systems,namely general
(GCN4 dependent) and basal (11, 183).
Whereas the general control promoter is active under con-ditions
of amino acid starvation, the basal control promoteris not
regulated by amino acids and is responsible for thehigh basal level
of transcription of the amino acid biosyn-thetic genes, even when
amino acids are present in thegrowth medium. There are some genes
for which, in theabsence ofGCN4 protein, the basal promoter is also
affectedand which therefore depend on GCN4 protein for at least
onecomponent of their basal expression (see below).
Transcriptional regulation of a yeast RNA polymerase IIpromoter
requires three kinds of cis-acting sequences,namely upstream, TATA,
and initiator elements (reviewedin references 185 and 186).
(i) Upstream elements (or upstream activation sequences[UASs])
are target sites for various activator proteins; theywork in a
distance- and orientation-independent mannerapproximately 100 to
600 bp upstream from the transcriptioninitiation site (75, 185,
186). In many respects, upstreamelements resemble enhancer elements
of higher eukaryotes.Genes subject to a common control mechanism
contain
upstream elements that are in general similar in the DNAsequence
that allows the binding of the same activatorprotein (for reviews,
see references 73 and 74). For thegeneral and basal control
promoters in yeast amino acidbiosynthetic genes, different upstream
elements exist asbinding sites for the various regulatory proteins
controllingthe basal or general control response.The optimal
promoter-binding site for the general control
regulator GCN4 is the well-characterized palindrome
5'-ATGA(C/G)TCAT-3' (10, 56, 83, 90, 94, 95, 148, 182). SuchGCN4
recognition elements (GCREs) have been found re-peated upstream of
every analyzed structural gene subject togeneral control (reviewed
in reference 87). The naturallyoccurring sites analyzed so far are
not identical to theconsensus sequence, but differ by 1 to 2 bp
(186). The GCN4protein binds general control promoters at all GCRE
se-quences (10). Deletion analysis of a number of these promot-ers
has demonstrated that GCRE sequences are both neces-sary and
sufficient for general control-mediated regulation oftranscription
in vivo (56, 181) and are therefore a class ofupstream activation
sequences (UAS). Little is known,however, about the interplay of
multiple GCREs in a natu-rally occurring general control promoter
in vivo. Somefeatures of the analyzed GCN4-regulated promoters of
thearomatic amino acid biosynthetic genes are summarized inTable
4.A similar sequence, TGACTA, contained in the recogni-
tion element for the mammalian transcription factor AP-1,can
interact with the yeast AP-1 homolog, yAP-1, a factor ofunknown
function, and stimulate transcriptional activationindependently of
GCN4 (78).For the basal-level control of the histidine
biosynthetic
gene HIS4, the trans-acting factors BAS1 and BAS2 havebeen
identified (11). The BAS2 gene, which is identical toPH02 and
GRFIO, is repressed in its expression by phos-phate and is
autoregulated (146, 210, 211). PHO2-BAS2binds to upstream elements
of HIS4, the acid phosphatasegene PHO5, and the aromatic amino acid
gene TRP4 pro-moter (11, 26, 193, 200) (see below).For the HIS3
gene it has been shown that upstream
elements necessary for basal gene expression are poly(dA-dT)
sequences, and it has been proposed that these act byexcluding
nucleosomes (187). An oligo(dA-dT)-binding pro-tein might be
involved in this kind of basal gene expression.Such a protein of
248 residues, named datin, has beenpurified and requires at least 9
to 11 bp of oligo(dA-dT) for
356 BRAUS
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 357
TABLE 4. Features of analyzed GCN4-regulated promoters of
several aromatic amino acid biosynthetic genes of S. cerevisiae
GCRE Initiation sites of transcriptionGene Position and Orienta-
In vitro GCN4
sqeca tin bnig Demonstrated function(s) Basaldenetsequence" tion
binding dependentAR03 -180 GTGACTAAT -188 a + UASGCb, basal control
-38, -31, -22, -8 As basalAR04 -312 ATGACTCAA -304 - + UASGC -136,
-88, -72 As basalTRP2 -162 CTGACTCAT -155 0 NDC -95, -59, -52,
-49,
-33TRP3 -162 TTGACTCAT -155 - ND -88, -80, -79, -77,
-49, -23, -21TRP4 -246 ATGACTAAT -238 (GCRE1) - + UASGC,
PHO2-binding site -127, -76 -31, -26,
-12-166 TTGACTCTC -158 (GCRE2) > + Together with GCRE3:
UASGC,
TATA box analog-151 ATGATTCAT -143 (GCRE3) - + Together with
GCRE2: UASGC,
TATA box analogTRP5 -233 GTGACTGGT -155 >0 ND Minor UASGC
-45, -28, -18
-108 ATGACTAAT -100 -a ND Major UASGCaMismatches with the
consensus sequence are indicated by bold letters.b UASGC, UAS sites
functional in general control derepression.c ND, not
determined.
high-affinity DNA binding (205). The exact function of
datinremains to be examined.The GCN4 protein can, in specific
cases, also regulate the
basal expression of amino acid biosynthetic genes, as shownfor
several genes, including AR03 (155) (see below), HIS4(11), and LEU2
(30).
(ii) As in mammalian promoters, TATA elements arenecessary but
not sufficient for accurate initiation of tran-scription in S.
cerevisiae (60, 185). TATA elements arelocated close to mRNA
initiation sites and mediate the firststep in the pathway of
transcription initiation by binding thegeneral transcription factor
TFIID (31, 32, 199). The yeasttranscription factor TFIID is able to
substitute for thecorresponding HeLa cell TATA-binding protein
(39). Incontrast to higher eukaryotes, in S. cerevisiae the
distancebetween TATA element and mRNA initiation site can
varybetween 40 and 120 bp (41). For the HIS3 promoter there
artdifferent TATA elements for the GCN4-dependent promoter(TATAR)
and the basal promoter (TATAC). For the regula-tory TATA element
(TATAR) in the HIS3 promoter, asaturation mutagenesis experiment
has been carried out, andit appears that only the sequences TATAAA
and (to a lesserextent) TATGTA or TATATA are functional in vivo
(42).Functional TATA elements are located between UAS andmRNA
initiation site(s) (187).Three possible models have been proposed
to explain how
specific activator proteins could interact with the
basictranscription machinery. In one model the specific
activatorinteracts with the TATA-binding factor TFIID to
facilitateassembly of a preinitiation complex. The assembled
preini-tiation complex would then interact with and activate
RNApolymerase II. In another model the activator would performsome
step after the assembly of the general factors into apreinitiation
complex (31), e.g., a direct interaction of theactivator with RNA
polymerase II (7, 14, 24). In a thirdmodel an additional protein,
termed an adaptor or mediator,is necessary to interact with the
specific activator, TFIID,and with RNA polymerase II (17, 102).
(iii) The transcription initiator element is the
primarydeterminant of the location where transcription begins in
S.cerevisiae (for a review, see reference 186). Yeast
mRNAinitiation sites are determined primarily by specific
initiator
sequences, not by the distance from the TATA element as inmany
genes of higher eukaryotes (41). An initiator as atranscription
control element is also described for the lym-phocyte-specific
terminal deoxynucleotidyltransferase gene(176). Two types of start
site selection patterns have beenfound in S. cerevisiae
GCN4-controlled genes when tran-scription start sites of the basal
expression were comparedwith the start sites of the GCN4-driven
transcription (140).Only a single start site of transcription has
been found in theHIS4 promoter region when the 5' ends of basal
controlledtranscripts, as well as GCN4-controlled transcripts,
weredetermined (140). The HIS3 promoter initiates
transcriptionequally from two sites, at + 1 and + 12, during basal
expres-sion. The GCN4-driven transcription of this promoter
thenpreferentially initiates at the basal initiation site at + 12
(41).
In the aromatic amino acid pathway of S. cerevisiae, fourof the
five TRP genes (132), the isogenes AR03 and AR04(190), and the AR02
gene (100a) are derepressed under thegeneral control system. The
genes TRPI, AR07, and TYR]are not derepressible by this system (28,
121, 169, 170). Theregulation of the other genes in the pathway
remains to beinvestigated. The arrangement of many of the
elementsdescribed above can be compared, as the promoters of
allfive TRP genes, the TYR] gene, and the AR03, AR04, andAR07 genes
have been cloned and sequenced (2, 64, 121,154, 170, 194, 212,
213). Although little is known about thebasal control aspect of
these promoters, more and more dataabout the GCN4-regulated parts
of the promoters are avail-able. The following section summarizes
the available datafor several of these promoters, with the main
focus on theGCN4-mediated regulation (Table 4).
(i) A single GCN4-binding site has different functions in
thepromoters of the isogenes AR03 and AR04. Both isogenes(AR03 and
AR04) encoding DAHP synthases in S. cerevi-siae respond equally
well to the general control regulatorysystem. In fact, DAHP
synthase activity can be increasedsixfold under derepressing
conditions, whereas, for exam-ple, TRP-encoded enzymes can be
derepressed only two- tothreefold (132, 190). Cells carrying only
one intact isogeneare phenotypically indistinguishable from a
wild-type strainwhen grown on minimal medium.
In contrast to AR04 and to other genes of the pathway (28,
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
143), a drop in ARO3 enzyme activity is observed in a
gcn4background. Concomitantly, the growth rate of an AR03aro4 gcn4
strain is reduced by 50%. Growth and enzymelevel can be restored by
transforming the mutant strain withthe GCN4 gene on a low-copy
plasmid, imitating a wild-typesituation. The complete functional
AR03 promoter com-prises 231 bp and contains a -180 GTGACTAAT
-188binding site for GCN4 in an inverse orientation (154) (Table4).
This corresponds to a 2-bp mismatch with the optimalpalindromic
binding site ATGA(C/G)TCAT. Mutating theAR03 element to GTTACTAAT
inhibits the binding ofGCN4 and results in the same phenotype as
that of the AR03aro4 gcn4 strain, namely a decreased basal level of
AR03gene product and slow growth of a strain defective in
itsisogene, AR04. In addition, AR03 gene expression cannotbe
increased under conditions of amino acid starvation (155).The
amount of GCN4 protein present in repressed wild-typecells
therefore seems to contribute to a basal level of AR03gene
expression.As found for AR03 and a few other genes, including
ILV2
and ARG4 (87), only a single GCN4-dependent UAS is foundin the
AR04 promoter (llla) (Table 4). This element, withthe sequence
5'-ATGACTCAA-3' in normal orientation (onemismatch to the consensus
sequence), is located at positions-312 to -304. A second identical
element was found ininverse orientation downstream of the AR04 ORF
at posi-tions + 1297 to + 1289, located only 185 bp downstream
ofthe translational stop codon. The two elements form aperfect 9-bp
inverted repeat with the coding sequence of theAR04 gene in
between.The upstream GCN4-binding site was shown to be the
upstream activation site of the AR04 gene, which is neces-sary
for GCN4-mediated transcription activation. Destroy-ing this
sequence does not affect the basal level of AR04expression, but
AR04 gene expression can no longer beincreased under amino acid
starvation conditions (155). Thesequence elements responsible for
the basal level of tran-scription have not yet been identified. The
GCN4-bindingsite located downstream of AR04 has no function
withrespect to the AR04 gene, but is a functional UAS of
anotheramino acid biosynthetic gene of histidine biosynthesis,HIS7,
located immediately downstream. This configurationdemonstrates one
of the differences between UASs andmammalian enhancers (llla).
Mammalian enhancers alsofunction when located downstream of the
gene, whereas aGCN4 site which is actually used in vivo is not able
to do so.The general control activator GCN4 thus has two func-
tions for these isogenes: (i) to maintain a basal level
ofAR03transcription (basal control) in the presence of amino
acidsand (ii) to derepress the AR03 as well as the AR04 gene toa
higher transcription rate under amino acid starvationconditions
(general control).Both promoters contain multiple initiation sites
of tran-
scription (Table 4). For AR03, four major 5' ends weremapped
between positions -38 and -8 upstream of the ATGstart codon (154).
For AR04 the three major transcriptioninitiation sites were
localized further upstream of the trans-lational start sites at
positions -136, -88, and -72. Tran-scripts starting from all
initiation sites are equally elevatedunder conditions of amino acid
starvation by the generalcontrol system (llla, 154).The four AR04
transcript ends were mapped 12 to 84
nucleotides upstream of the HIS7 upstream element, sug-gesting
that there is virtually no intergenic space betweentranscription
termination and promoter elements of thesetwo genes (llla).
(ii) TRP2 and TRP3 promoters. The products of the twogenes TRP2
and TRP3 form a heterodimeric enzyme com-plex which consists of
equimolar amounts of both polypep-tide chains (160). Therefore,
expression of the two genesmust be coordinated. In both genes a
GCN4 consensussequence with a single mismatch is located in the
promoter:at position -162 CTGACTCAT -155 for TRP2 and -124TTGACTCAT
-116 for TRP3 (Table 4). Several transcrip-tion start sites were
mapped in both promoters (-95, -59,-52, -49, and -33 for TRP2; -88,
-80, -79, -77, -49,-23, and -21 for TRP3 (212). A comparison of the
strengthsof the two promoters has still not been undertaken.
(iii) Three GCN4-responsive elements have different func-tions
in the TRP4 promoter. The promoter of the TRP4 geneof S.
cerevisiae, coding for the enzyme anthranilate
phos-phoribosyltransferase (64) contains two putative UAS ele-ments
for the GCN4 protein. UAS1 comprises a singleGCN4-binding site -246
(relative to the translational startsite) ATGACTAAT -238,
designated as GCRE1 (one mis-match), and UAS2 comprises two
adjacent repeats, -166TTGACTCTC -158 and -151 ATGATTCAT -143,
desig-nated as GCRE2 (three mismatches) and GCRE3 (one mis-match),
respectively. UAS1 and UAS2 are both able tospecifically bind the
activator protein GCN4 in vitro (26)(Fig. 2; Table 4).
All three GCREs are required for a normal GCN4-depen-dent
transcription activation but do not affect basal tran-scription. A
promoter containing a mutation of either UAS1(gcrel) or UAS2
(gcre2-gcre3) is no longer inducible by theGCN4 protein (134). The
use of the TRP4 promoter byGCN4 is reduced to approximately 30%
when either GCRE2or GCRE3 is mutated (134).GCN4 has been shown to
compete at the UAS1 site with
another transcriptional regulator, PHO2/BAS2 (26). PHO2/BAS2
encodes a homeo-box protein (33, 173) homologous togenes involved
in developmental regulation in many differentspecies (65). Among
other functions, PHO2/BAS2 appearsto be closely involved in
regulating Pi metabolism.PHO2/BAS2 binds directly to the PHO5,
HIS4, and TRP4
promoters. In the TRP4 promoter the PHO2/BAS2-pro-tected region
in vitro comprises approximately 20 nucleo-tides and completely
overlaps the GCN4-protected UAS1region. GCN4 and PHO2/BAS2 bind to
UAS1 in a mutuallyexclusive manner. PHO2/BAS2 does not affect the
basallevel of TRP4 expression, indicating that additional cis-
ortrans-acting factors are involved in basal TRP4 expression.When
PHO2/BAS2 competes with GCN4 at the UAS1 site ofthe TRP4 promoter,
it prevents TRP4 derepession underconditions of simultaneous Pi and
amino acid starvation (26).Whereas GCN4 mediates the response of
the transcriptionalapparatus to the environmental signal amino acid
limitation,PHO2/BAS2 could be the phosphate sensor that adjusts
theresponse to the availability of phosphate precursors
fortryptophan biosynthesis. The physiological significance ofthis
is apparent when it is considered that TRP4 encodes
aphosphoribosyltransferase, requiring 5-phosphoribosyl
1-py-rophosphate (PRPP) as a substrate. Therefore, repression ofthe
GCN4-induced TRP4 expression prevents more enzymefrom being
produced under conditions where PRPP as one ofthe substrates is
limiting.The mode of action of PHO2/BAS2 seems to differ de-
pending on the context of the binding site of the correspond-ing
target genes: PHO2/BAS2 and PHO4 are both necessaryfor PHOS and
PHOIJ activation under conditions of phos-phate starvation (151,
200, 209-211). In addition, PHO2/BAS2 has another function in
another amino acid promoter.
358 BRAUS
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 359
10bp
basal control-mediated transcription
GCRE1 GCRE2 GCRE3
UASI
IUAS2
GCN4-mediated transcription:
GCN4
GCN4
PH02GCN4
GCN4GCN4
GCN4TFIID
+4.
+4.
FIG. 2. TRP4 promoter of S. cerevisiae. The effects of the
binding or nonbinding of different combinations of the
transcription factorsGCN4, PHO2, or TFIID to UAS1 and UAS2 on the
formation of GCN4-mediated transcripts (starting at initiation
sites -31 [i31], -26 [i26],and -12 [i12] relative to the
translational start codon) are shown. The expression of the basal
control-mediated transcripts (starting atinitiation sites -127
[i127] and -76 [i76]) is not affected under these conditions. See
the text for details.
Together with BAS1, PH02/BAS2 is necessary for basal-level
expression of the amino acid biosynthetic gene HIS4(11). HIS4
encodes a trifunctional histidine biosyntheticenzyme. The enzyme
preceding the HIS4 gene product is aphosphoribosyltransferase
(encoded by the HIS] gene) thatalso requires PRPP as a substrate
(101). It is still unclearwhether the PH02/BAS2 protein actually
functions as aPRPP sensor in the cell.
Basal transcription and GCN4-mediated transcription ini-tiate at
different start sites at the TRP4 promoter. A basallevel of TRP4
transcription results in transcripts starting attwo sites at
positions -127 (i127) and -76 (i76) relative tothe translational
start site. Under conditions of derepressionby GCN4, the basal
transcripts remain unchanged but threeadditional signals for mRNA
start sites appear at positions-31, -26, and -12 (63, 134); these
were named i31, i26, andi12 (Fig. 2). These additional transcripts
correspond to theincrease in transcription initiation as measured
at the mRNAand enzyme levels and therefore represent the product of
theGCN4-driven part of the TRP4 promoter. These GCN4-dependent
start sites are lacking when the GCN4 regulator ismissing from the
cell as well as when the GCN4-driventranscription of the TRP4 gene
is abolished by mutations inUAS1 or UAS2 or both. The use of the
initiator elements i31,i26, and i12 by the transcription machinery
is thereforesolely dependent on the presence of the regulator
proteinGCN4 and its recognition elements in the TRP4 promoter.These
results show that basal transcription and GCN4-driven transcription
of the TRP4 gene are distinct events,even with respect to their
transcription start sites.Upstream activator proteins such as GCN4
or GAL4
normally stimulate transcription when bound upstream of aTATA
element. No functional consensus TATA box (likeTATAAA, TATATA or
TATCTA [42]) is found in the TRP4promoter between UAS2 and the
transcription initiation
sites of the GCN4-mediated transcription (63). This obser-vation
led to the question of whether UAS2 is the analog ofa TATA box for
the GCN4-dependent TRP4 promoter. Toanalyze this question, the
TRP4-UAS2 element was ex-changed for a consensus TATA box, TATAAA,
which isidentical to the GCN4-dependent TATA element in the
HIS3promoter (42) and to the CYCI-52 TATA element that bindsto the
transcription factor TFIID in vitro (76). Expressionstudies
revealed that the newly introduced TATA box wasable to restore the
GCN4-driven transcription of a TRP4promoter with a mutated UAS2.
The basal level of TRP4transcription was unaffected; transcription
of the mutantTRP4 promoter started mainly at i127 at repressed
levels orin the absence of GCN4 protein, as was found for
thewild-type promoter. Transcription initiated again at i31,
i26,and i12 at high levels of GCN4 protein in the cell.
Theregulated initiator elements i31, i26, and i12 can therefore
beused in two possible ways: (i) when transcription is driven
byGCN4 acting synergistically via UAS1 and UAS2
(wild-typesituation) and (ii) when transcription is dependent on
GCN4binding at UAS1 and on a TATA factor (presumably TFIID)binding
to a TATA box situated at the position of UAS2.These results show
that a consensus TATA box can func-tionally replace the UAS2
element in the GCN4-dependentTRP4 promoter, suggesting that the
UAS2 element has afunction in vivo which is analogous to that of a
TATAelement in other eukaryotic promoters. A possibility
whichcannot completely be ruled out is that, in vivo, the threeGCN4
sites serve as UASs that activate transcription incombination with
a more downstream weak TATA element,which deviates somewhat from a
TATAAA sequence. Inaddition, other factors with binding properties
similar toGCN4, e.g., the transcriptional factor yAP-1 (78), might
beinvolved in the function of UAS2. There is, however,additional
evidence that the TATA factor function for the
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
general control transcription in the TRP4 gene is fulfilled
byGCN4. (i) GCN4 is able to interact specifically with
RNApolymerase II in vitro (24). The region of GCN4 thatcontacts
polymerase II resides within the DNA-bindingdomain of the protein
and not the short acidic domain, whichis required for
transcriptional activation in vivo. (ii) GCN4efficiently activates
transcription in an artificial GAL-HIS3hybrid promoter in the
absence of a TATA element, whenbound close to the mRNA initiation
site (43). These datademonstrate that there are other factors in S.
cerevisiae,apart from the general transcription factor TFIID,
thatrecognize sequences unrelated to the consensus TATA boxbut are
nevertheless able to perform the role of TFIID.Taken together,
these results suggest that GCN4 can
activate transcription by exhibiting two alternative
functionswithin one natural promoter (Fig. 2).
(iv) Two putative GCN4 elements in the TRP5 promoter.The TRP5
promoter contains a putative GCN4-binding site-108 ATGACTAAT -100
relative to the translational startcodon, which contains one
mismatch in comparison with theconsensus sequence (213). Deletion
of this sequence abol-ishes the general control response (135,
136). A secondsequence -233 GTGACTGGT -225 contains three
mis-matches and seems to be necessary for full derepression andalso
for basal-level expression (Table 4). In addition, othersequences
in this region seem to be important for highbasal-level expression
(136). Three different transcriptionalstart sites, located at
positions -45, -28 and -18, have beenidentified (213).
(v) The promoters of AR07, TRPI, and TYRI are notregulated by
GCN4. Although the general control systemregulates most of the
genes of the aromatic amino acidpathway, it does not seem to be
necessary to regulate allbiosynthetic genes in order to maintain
the metabolic flowthrough the pathway. No GCN4 regulation was found
forone gene of the tryptophan branch, TRPJ (27, 28), and onegene of
the tyrosine branch, TYRJ (121). For the AR07 gene,encoding
chorismate mutase, neither transcriptional regula-tion by the
general control system nor specific regulation byaromatic amino
acids was found (169, 170).
Similar to the promoters of other aromatic amino
acidbiosynthetic genes, however, a recognition element for theGCN4
transcriptional activator of amino acid biosynthesis ininverse
orientation is located 425 bp upstream of the firsttranscriptional
start point in the AR07 promoter (-496ATGACTGAA -504; two
mismatches with the consensussequence). This element binds GCN4
specifically in vitro.Northern (RNA) analysis and determination of
the specificenzyme activity reveal that the element is not
sufficient tomediate transcriptional regulation by GCN4 in vivo
(169).These data suggest that in addition to a consensus
sequencecapable of binding the GCN4 protein, other
DNA-bindingproteins or other parameters, such as chromatin
structure,determine whether a recognition site for- a
transcriptionfactor functions as a UAS. For the AR07 mRNA, three
5'ends at positions -36, -56, and -73 relative to the startcodon
were mapped.TYR], the gene which encodes prephenate dehydroge-
nase, is also not regulated by the GCN4 system, and there isno
consensus sequence for a GCN4-binding site located inthe promoter.
Instead, transcriptional regulation seems to bedependent on the
presence or absence of phenylalanine infusions between the TYRJ
promoter and the CAT (chloram-phenicol acetyltransferase) reporter
gene. Only a singletranscriptional start site was found in the TYRJ
promoter, atposition -70 relative to the translational start codon
(121).
In the TRPI promoter the only site similar to a GCN4-binding
site (-54 CTGACTATT -46) has three mismatches(194). This sequence
is located between the transcriptionstart sites (27) and is unable
to bind GCN4 in vitro (94).Transcription from the TRPI promoter is
initiated onlyapproximately half as frequently as, for example,
transcrip-tion from the TRP3 promoter (28). The TRPI
promotergenerates two groups of transcripts (103) corresponding
tofive transcription initiation sites, organized in two
clusters.The two longer transcripts start at positions -209 and
-187,and the three shorter transcripts start at positions -36,
-26,and -16. A transcriptional terminator element of
unknownfunction located in the 5' region upstream of the
TRPIpromoter seems to be essential for accurate TRPI
geneexpression. In partial TRPI promoters-lacking the
termi-nator-transcription is initiated predominantly in
adjacentupstream regions, resulting mainly in large, poorly
translatedtranscripts. The effect can be suppressed by
introducingartificial transcription barriers such as
transcriptional termi-nators, centromere sequences, or yeast
replicator (ARS)sequences in front of the truncated TRPI promoter
(27). Inaddition, an A+T-rich region of dyad symmetry was pro-posed
as a promoter element for the shorter transcripts(104). This
element consists of two perfect inverted repeatsof 12 A+T rich
nucleotides separated by a 21-bp spacer andlocated between
positions -81 and -125 upstream of thestart codon. Deletions within
this element abolished tran-scription of the shorter transcripts
(104).mRNA decay. mRNA decay is a potential control point of
gene expression (for a review, see reference 29). In
aromaticamino acid biosynthesis, differences in mRNA
stabilityaffect the relative steady-state level ofmRNAs in at least
onecase. For three mRNAs of the aromatic amino acid biosyn-thetic
pathway, the half-life was determined. For the largerTRP3 and TRP4
mRNAs the half-lives were determined as11 and 14 min, respectively
(28, 63). For the smaller TRPItranscript the half-life was 19 min,
indicating a greaterstability (28). Therefore, TRPI mRNA is
approximatelytwice as stable as TRP3 mRNA. Since transcription from
theTRPI promoter initiates only approximately half as fre-quently
as from the TRP3 promoter, the final steady-stateamounts of the two
mRNAs without amino acid limitationare similar (28).
Translation
There is no evidence that translation of the structuralgenes of
aromatic amino acid biosynthesis plays a major rolein the
regulation of gene expression. The expression of themain
transcriptional regulator of the pathway, GCN4, how-ever, which is
the basis of the regulation of gene expressionof most of the
structural genes of the pathway, is regulatedby amino acid
availability at the translational level.The transcriptional
regulator of amino acid biosynthesis
GCN4 is regulated at the translational level. The
translationalderepression of GCN4 mRNA seems to be directly linked
tomajor changes in the protein synthesis machinery of the
cell(197). The GCN4 mRNA has a 600-bp leader sequencecontaining
four short ORFs, each consisting of an AUGcodon followed by one or
two sense codons and then atermination codon (87, 191). This
sequence organization isunusual because in S. cerevisiae, like in
other eukaryotes,the AUG codon most proximal to the 5' end of the
mRNA ingeneral functions as the translational initiation signal.
Addi-tional AUG codons upstream of the normal translationinitiation
substantially inhibit translation of downstream
360 BRAUS
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 361
coding sequences (106, 107). Under certain circumstancesthe 40S
ribosomal subunit can resume scanning as soon asribosomes terminate
translation of an upstream ORF and canreinitiate to a certain
degree at a downstream ORF (107, 108,157, 198). Removal of the four
upstream ORFs from theGCN4 transcript, either by deletion or by
point mutations inthe four ATG start codons of the corresponding
DNAsegment, results in high-level unregulated GCN4 expression.In
addition, insertion of the four upstream ORFs into theleader
sequence of another transcript results in translationalregulation
of gene expression similar to that of the GCN4transcript itself. A
combination of the first and the fourth ofthese upstream ORFS is
sufficient for wild-type regulation ofGCN4. Whereas the fourth
upstream ORF seems to be thestrongest translational barrier for
GCN4 expression, the firstupstream ORF is rather leaky and seems to
be a positiveelement to bypass ORF4 under amino acid starvation
con-ditions (138, 196). In amino acid-starved cells the
ribosomalreinitiation at the upstream ORF seems to be
suppressed.The 40S ribosomal subunits seem to traverse the
upstreamORFs and to ignore their AUG start codons (1). The
inhib-itory effect of ORF4 on GCN4 expression seems to bedetermined
mainly by sequences surrounding its stop codon(131). Different
trans-acting regulators are involved in thefunction of the four
upstream ORFs (for reviews, see refer-ences 87 and 88).Two classes
of mutations, gcn and gcd, have been de-
scribed conferring either a nonderepressible (GCN) or
aconstitutively derepressed (GCD) phenotype of enzymesregulated
under general control. From genetic studies, acascade model was
proposed to explain the interactionsbetween the various genes and
gene products (86, 89). In thishierarchy the GCN4 gene product was
identified as the mostdirect activator of the structural genes
subject to generalcontrol (87, 191).According to this model, the
GCD genes are negative
effectors of GCN4 translation. GCDJ and GCD2 are essen-tial
yeast genes; they have been cloned and sequenced, buttheir exact
function is unknown (84, 152, 153). Among thesenegative effectors
there are a- and ,-subunits, respectively,of the eukaryotic
initiation factor 2 encoded by the genusSUI2 and SUI3 (88, 204).
The GCN genes GCNI, GCN2,and GCN3 are assumed to be positive
regulators whichantagonize the negative effect of the GCD-encoded
regula-tors on GCN4 expression (77, 152).The translational
derepression of GCN4 mRNA depends
on the small ORFs in the leader region and on positiveeffectors
such as GCN2 and GCN3 (86, 138). One domain ofthe GCN2 gene product
is a protein kinase that is posttrans-lationally regulated. Its
substrate specificity in vivo is notknown (165, 202). There is an
additional domain with homol-ogy to histidyl-tRNA synthetase (201).
Although depletion ofan amino acid pool leads to general control
derepression, areduction in the level of tRNA aminoacylation seems
to bethe more direct signal for derepression (126, 177a).
Sinceaminoacyl-tRNA synthetases bind uncharged tRNA as asubstrate
and are able to distinguish between charged anduncharged tRNAs, it
has been proposed that the putativehistidyl-tRNA synthetase domain
of GCN2 monitors theconcentration of uncharged tRNAs in the cell
and activatesthe adjacent protein kinase domain under starvation
condi-tions when uncharged tRNA accumulates (88, 201). It hasbeen
shown that translational activation of GCN4 can betriggered in a
cell-free system by uncharged tRNAs (111).There is evidence that
increased levels of the GCN2
protein kinase in the cell increase the ability of 40S ribo-
somal subunits that have participated in the translation of
thefirst of the four small ORFs (ORF1) to reinitiate at adownstream
AUG (198). Whatever the function of the GCN2protein kinase in this
process, it can be implemented only inconjunction with the
translation of ORF1 (198). Moreover,the 5' region of the GCN2 gene
contains a recognition site,which binds the GCN4 protein in vitro
(165). It is still anopen question whether there exists a
transcriptional-transla-tional circuit which involves the GCN4
transcriptional acti-vator and the GCN2 protein kinase (165, 198).
The precisemechanism of the mode of action of the GCN2 protein
kinaseand the other trans-acting factors (GCN and GCD
geneproducts), modulating the expression of GCN4 directly
orindirectly, is a topic of current research.
REGULATION OF ENZYME ACTIVITY
In S. cerevisiae the regulation of enzyme activities plays
amajor role in the establishment of the flow through an aminoacid
biosynthetic pathway. For the biosynthesis of aromaticamino acids,
the main control points are the entrance of theshikimate pathway
and the first branch point, where thedistribution of chorismate
either in the direction of tryp-tophan or in the direction of
phenylalanine-tyrosine is con-trolled. The effector molecules are
the end products of thepathway (Fig. 1).
In S. cerevisiae the initial step of the pathway catalyzed bythe
two DAHP synthases is regulated by feedback
inhibition;phenylalanine feedback inhibits the ARO3-encoded
DAHPsynthase, and tyrosine feedback inhibits the ARO4-encodedDAHP
synthase. A tryptophan-inhibitable DAHP synthasehas not been found.
This situation leads to tryptophanstarvation for yeast cells grown
in a medium lacking tryp-tophan but with an excess of phenylalanine
and tyrosine andcan be compensated by increasing the rate of
enzymesynthesis with the general control system. Mutant
yeaststrains which are unable to elevate transcription by
theGCN4-dependent general control system have a
significantlyreduced growth rate under these conditions (144). At
the firstbranch point the tryptophan biosynthetic anthranilate
syn-thase complex (encoded by TRP2-TRP3) is feedback inhib-ited by
tryptophan and the tyrosine-phenylalanine biosyn-thetic chorismate
mutase (encoded by ARO7) is feedbackinhibited by tyrosine. In
addition to feedback inhibition, thechorismate mutase is also
strongly activated by tryptophan,the end product of the other
branch. This dual control ofenzyme regulation by tyrosine as
feedback inhibitor andtryptophan as activator is unique when
compared with thecorresponding enzymes of other organisms.The
following section focuses mainly on the regulation of
the activities of these enzymes.
Phenylalanine-Inhibitable DAHP SynthaseTakahashi and Chan (189)
were able to separate and
preliminarily characterize the DAHP synthase isoenzymesby
affinity chromatography. By using current gene technol-ogy, the
problem of separating two isoenzymes was circum-vented for the
phenylalanine-inhibitable DAHP synthase byoverexpressing the AR03
gene on a high-copy-number plas-mid in a yeast strain carrying an
aro4 deletion (156). Thephenylalanine-inhibitable DAHP synthase
from S. cerevisiaehas been purified to apparent homogeneity by a
1,250-foldenrichment of the enzyme activity present in wild-type
crudeextracts, by using the overproducing strain. Hence thisDAHP
synthase corresponds to approximately 0.1% of the
VOL. 55, 1991
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
MICROBIOL. REV.
TABLE 5. Features of branch point enzymes for the biosynthesis
of aromatic amino acids in S. cerevisiae
eOligomericMolecular mass (S)0%5or Km ka) KJK. % of totalEnzyme
Gene statoeri of native (SM)orSK-k') K(/K) cellularenzyme (kDa)
(m)(1 m)protein
Phe-inhibitable DAHP syn- AR03 Monomer 42 For PEP, 0.018; for 10
For Phe/E4P, 0.01 0.05-0.1thase (EC 4.1.2.15) E4P; 0.130
Chorismate mutase AR07 Homodimer 60 Without aaa, 4.0 176 0.01(EC
5.4.99.5) + Tyr, 8.6 129 For Tyr, 0.05
+ Trp, 1.2 264 For Trp, 0.0015Gln-dependent anthranilate TRP213
Heterodimer 130 For chorismate, 0.05; ND For Trp, 0.056 0.05
synthase/InGP synthase for CDRP, 0.25(EC 4.1.3.27/EC
4.1.1.48)
2PRA isomerasec TRPI Monomer 23 For PRA, 0.005 50 0.006
aaa, amino acid.b ND, not determined.c PRA isomerase is included
for comparison, because in other fungi this enzyme activity is part
of a trifunctional enzyme; Gln-dependent anthranilate
synthase-InGPsynthase-PRA isomerase.
total cellular protein. This is in agreement with an
estimationof the AR03 mRNA level in the cell, which, on the basis
ofthe codon usage index of Bennetzen and Hall (15), amountsto
roughly 0.05% of the total mRNA.
Gel filtration indicates that the active enzyme is a mono-mer
with a molecular mass of 42 kDa, which corresponds tothe calculated
molecular mass deduced from the previouslydetermined primary
sequence (154). A comparison with thethree described DAHP synthases
from E. coli reveals strik-ing differences in the quaternary
structure: the phenylala-nine-inhibitable enzyme is a tetramer,
whereas the tyrosine-inhibitable enzyme is a dimer (124, 171, 175),
although thereis 70% sequence similarity between the enzymes (52,
174).The tryptophan-inhibitable DAHP synthase has been de-scribed
as a tetramer (145). Another monomeric DAHPsynthase as in S.
cerevisiae has as yet not been described.Atomic absorption
spectroscopy suggests that the yeast
phenylalanine-inhibitable DAHP synthase is an iron
metal-loenzyme. The enzyme can be inactivated by EDTA in areaction
which can be reversed by the addition of severalbivalent metal
ions. Similar results have been found for
thephenylalanine-inhibitable E. coli enzyme (123, 175, 179).The
kinetic data of the phenylalanine-inhibitable yeast
DAHP synthase (summarized in Table 5), with a calculatedrate
constant of 10 s-1 (156), suggest a sequential reactionmechanism
similar to that proposed for the tyrosine-in-hibitable E. coli DAHP
synthase (171). The apparent Mi-chaelis constant of the enzyme is
0.018 mM for PEP, whichis similar to the values obtained for the
two available DAHPsynthase isoenzymes from E. coli and the
tryptophan-sensi-tive enzyme from N. crassa (124, 145, 171, 175).
For E4P theMichaelis constant is 0.13 mM, and the reported values
forthe other described enzymes range between 0.0027 and 0.9mM. A
reason for this finding may be that E4P forms dimersin solution,
but the enzyme distinguishes between monomersand dimers (145), so
that the concentration of this substrateavailable to the enzyme
could be overestimated.The reported inhibition constants for all
the DAHP syn-
thases are of the same order of magnitude. For the yeastenzyme,
inhibition by phenylalanine is competitive withrespect to E4P and
noncompetitive with respect to PEP,with a Ki of 0.01 mM (156). The
N. crassa enzyme shows thesame pattern of inhibition, but only
qualitative results wereobtained, as the inhibition by tryptophan
was not hyperbolicand the intercept and slope replots curved
upwardly (145).
The present knowledge suggests that the interplay of
DAHPsynthase subunits and the regulatory behavior seems to
bedifferent in various organisms, although the high degree
ofsequence homology points to similarities in catalytic
behav-ior.Once the AR04 gene product is purified, it will be
inter-
esting to compare the quaternary structure and the kineticand
inhibitory properties of the two enzymes from S. cere-visiae.
The Allosteric Chorismate Mutase Can Be Locked in theActivated
State
The AR07 gene of S. cerevisiae encodes a
monofunctionalchorismate mutase catalyzing the first step in the
phenylal-anine-tyrosine branch. Interestingly, this Claisen
reactionfrom chorismate to prephenate can also be catalyzed
ste-reospecifically by a monoclonal antibody (85). Whereas
theallosteric chorismate mutase activity can be activated up
to10-fold in the presence of the specific effector tryptophan atthe
enzyme level, tyrosine is able to reduce the chorismateactivity up
to 10-fold, resulting in a range of regulation of theenzyme of a
factor of 100. No effect of phenylalanine, theother end product of
this branch of aromatic amino acidbiosynthesis, is known
(168).Mutant strains carrying the AR07C (constitutively acti-
vated chorismate mutase) alleles show increased sensitivityto
the amino acid analog 5-methyltryptophan with respect togrowth and
exhibit a 10-fold increase in the basal activity ofchorismate
mutase (110, 168, 170). The mutant enzymes arepractically
unresponsive to tyrosine and tryptophan. Sinceanthranilate synthase
and chorismate mutase control thedistribution of chorismate at the
first branch point of aro-matic amino acid biosynthesis, a high
chorismate mutaseactivity depletes the chorismate pool, destroys
the balancebetween the two enzymes and the chorismate pool in
thecell, and causes tryptophan starvation in the presence of
thefalse anthranilate synthase inhibitor 5-methyltryptophan(132,
160). Overexpression of the cloned mutant enzyme ona
high-copy-number vector leads to starvation for tryp-tophan because
of depletion of the chorismate pool even inthe absence of the
analog.
Recently a tryptophan auxotrophic mutant strain
ofPichiaguilliermondii was isolated that had a
sevenfold-increasedchorismate mutase activity, leading to a
depletion of the
362 BRAUS
on March 31, 2021 by guest
http://mm
br.asm.org/
Dow
nloaded from
http://mmbr.asm.org/
-
AROMATIC AMINO ACID BIOSYNTHESIS IN S. CEREVISIAE 363
chorismate pool (19). Once the cloned gene and the
purifiedenzyme are available, it will be interesting to compare
themwith the S. cerevisiae chorismate mutases.The wild-type and the
mutant chorismate mutases have
been purified approximately 11,000-fold from
overproducingstrains. On the basis, chorismate mutase represents
approx-imately 0.01% of the total cellular protein, and this
figurecorrelates with the estimated ARO7 mRNA level in the
cell,which, on the basis of the codon usage index of Bennetzenand
Hall (15), amounts to roughly 0.01%.The ARO7C phenotype is caused
by an identical point
mutation found in independent mutant alleles in the C-ter-minal
part of the 256-amino-acid protein. This mutationcauses a change
from threonine in the C-terminal part of thechorismate mutase at
amino acid 226. In a Chou and Fasmansecondary plot (44), the
replacement of the hydrophilicthreonine in the wild-type enzyme by
the hydrophobicisoleucine in the mutant enzyme interrupts a
hydrophilica-helical conformation from amino acids 220 to 226
(170).The wild-type and mutant enzymes are dimers consisting
of two identical subunits with a molecular mass of 30 kDa,each
one capable of binding one substrate and one activatormolecule
(tryptophan). Each subunit of the wild-type en-zyme also binds one
inhibitor molecule (tyrosine). Themutant enzyme which is still able
to bind tryptophan losesthe tyrosine-binding ability, suggesting
that there are twoconformational states for the wild-type enzyme
but only onefor the mutant enzyme (168).A 'H nuclear magnetic
resonance spectrum of the enzyme
reaction with the pure enzyme shows a direct and irrevers-ible
conversion of chorismate to prephenate without theaccumulation of
any enzyme-free intermediates. Since theother purified chorismate
mutases from E. coli are bifunc-tional and do not release
prephenate (50, 51), the use of thepurified yeast chorismate mutase
makes it possible to ob-serve the enzyme reaction for the first
time by 'H nuclearmagnetic resonance spectroscopy (168).
Table 5 summarizes some features of the wild-type cho-rismate
mutase. The kinetic data for